N3LO non singlet anomalous dimesions

[giacomomagni]

Main references:

• 1610.07477, nf^3,2 exact parts
• 1707.08315, Low moments. N->Inf limit. nf^0,1 leading nc parts.
• 2202.10362, N -> 0 limit

Plan:

• The nf^3 is given exact.
• The nf^2 is given exact, but already too complicated and contains some w=5 harmonics not available in eko. We try a parametrisation
• The nf^1, nf^0 is given exact in the large nc limit. Too complicated, it contains up to w=7 harmonics. Try a parametrisation.

[codecov]

Codecov Report

Merging #118 (7e9a258) into develop (1813ad7) will increase coverage by 43.08%.
The diff coverage is 100.00%.

@@             Coverage Diff              @@
##           develop     #118       +/-   ##
============================================
+ Coverage    52.34%   95.42%   +43.08%     
============================================
  Files           67       72        +5     
  Lines         3179     3301      +122     
============================================
+ Hits          1664     3150     +1486     
+ Misses        1515      151     -1364     

Flag	Coverage Δ
isobench	50.68% <0.00%> (-1.66%)	⬇️
unittests	95.42% <100.00%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/eko/harmonics/__init__.py	100.00% <ø> (+64.70%)	⬆️
src/eko/harmonics/g_functions.py	100.00% <ø> (+67.03%)	⬆️
src/eko/anomalous_dimensions/as4/__init__.py	100.00% <100.00%> (ø)
src/eko/anomalous_dimensions/as4/gnsm.py	100.00% <100.00%> (ø)
src/eko/anomalous_dimensions/as4/gnsp.py	100.00% <100.00%> (ø)
src/eko/anomalous_dimensions/as4/gnsv.py	100.00% <100.00%> (ø)
src/eko/harmonics/log_functions.py	100.00% <100.00%> (ø)
...eko/matching_conditions/operator_matrix_element.py	100.00% <100.00%> (+55.14%)	⬆️
src/eko/harmonics/constants.py	100.00% <0.00%> (ø)
src/eko/evolution_operator/grid.py	100.00% <0.00%> (+6.00%)	⬆️
... and 49 more

[alecandido]

Mostly comments on names and docs, but I guess it is expected: it is hard to comment on numbers, though I could comment on the code generating them.

Please add the Mathematica notebook in /extras/n3lo-ad (or something like this, choose your favorite name for the folder).

[alecandido]

I wonder if it is worth to repeat the references used function by function.

But maybe it is better like this, otherwise we bloat all docstrings with boilerplate. Just add a pointer (see :mod:'X' for more details) in order to make it easier to retrieve the source.

[alecandido]

@giacomomagni can you tell me how did you obtain the update in e5382d6?

Because in principle pyproject.toml is the same, so you're only getting compatible updates of deps and deps of deps, but in e5382d6 it happened that a considerable amount of deps switched from "dev" to "main" kind.
Can you tell me in particular the version of Poetry you used?

Just in case: you've done nothing wrong, I would not want to look aggressive. I'm only trying to monitor Poetry, and keep it under control (and in any case, reverting one commit is a breeze, so unless you delete a good bunch of them is hard you could cause any real problem).

[giacomomagni]

@giacomomagni can you tell me how did you obtain the update in e5382d6?

Maybe I messed up while merging develop into here. I saw here were conflicts in poetry.lock and since you updated it in 9207826, I accepted the current status. But then in 7e4e3a7 all the test were failing because of poetry, so I just did poetry update and commit as it was suggested.
I'm using Poetry version 1.1.12

[alecandido]

Last version of Poetry is v1.1.13, but I reran lockfile generation and obtained a very minor difference for a single package (not even worth to push).
Everything straight, the difference above should have been caused by nbsphinx.